2025: The Year AI Moves From Potential to Production
- By Davor Bonaci, DataStax
- January 18, 2025

Gen AI is poised to enter production at scale in 2025. While AI's potential has been clear for some time, achieving widespread production use has required more than technical readiness alone. Over the past year and a half, as an industry, we have found some significant technical challenges to running Gen AI at scale, which have been mainly solved.
The remaining hurdles relate to issues like how to design the best user experience, manage costs and deploy the technology effectively to users or customers, which are common to all IT projects. We also don’t yet have the defined regulatory frameworks in place that are settled enough. These issues will be addressed mainly in the coming months, positioning 2025 to see substantial growth in production-level AI.
Delays in production AI: Bridging technology and organizational readiness
While the core technology for production AI is ready, the people, processes, governance, and operational complexities have required more time. Technology alone isn’t sufficient; organizational dynamics, such as integrating AI into existing workflows and addressing regulatory concerns, tend to move slower. Additionally, many organizations have preferred a "fast follower" approach, allowing others to test and refine AI applications before committing fully themselves. This cautious stance is not due to a lack of capability but instead reflects a strategic choice to avoid perceived risks.
Navigating management frameworks to mitigate AI risks
Unintended consequences are likely, as the management frameworks needed for handling the complexities of AI will take years — possibly a decade — to fully mature. While companies aim to minimize AI biases and unexpected behaviors, these issues will still emerge. For example, an AI may inadvertently treat groups differently, like the case of a widely known AI assistant responding differently to presidential candidates, an outcome that wasn’t intended.
These unintended behaviors may often seem trivial, but in some instances, the impacts could be more serious. Ensuring AI acts responsibly will require more robust governance, which will take time to evolve across organizations and industries. We have to have that responsible AI approach in place where we can follow outcomes back to the AI model or to the training data to stop mistakes from happening again. GenAI is non-deterministic — you will not get the same result every time. So, we have to learn and keep optimizing for the most relevant responses.

Risks of over-complicating AI deployments
Fear and uncertainty, particularly in highly regulated industries, can lead companies to over-complicate their AI strategies or invest in areas that may not ultimately be necessary. While this may seem a misstep, adopting any new technology is natural. A cautious approach is often warranted in these sectors, where companies may prefer to over-invest initially rather than risk non-compliance or other issues. With technology that is still this new, you are making a bet on how well it will deliver, and then you can optimize your approach over time to increase your chances of success.
AI platforms: Easing user fears and ensuring responsible use
Many AI platforms address different parts of the stack. Depending on what they address, AI platforms can help reduce user fears by making it easier to understand how you build and manage AI successfully. Understanding the whole process makes you less likely to overlook essential elements or experience blind spots. Getting your tools so they are designed to work seamlessly together increases visibility and observability across AI operations, allowing you to monitor and understand AI behaviors better and respond appropriately. Having such oversight helps you feel more in control and reassured about the responsible operation of your AI solutions.
AI apps evolving: Beyond chatbots
Next year, we’ll likely see better chatbots and new AI use cases. Chatbots will become much more capable, evolving from merely answering questions and directing users elsewhere to perform tasks within the conversation. For example, instead of just telling you how to reset your password, future chatbots can help you reset the password directly within the chat interface.
Beyond chatbots, AI-driven information retrieval and synthesis play a major role. For example, instead of browsing through hundreds of product reviews on an eCommerce site about something you are interested in, AI will synthesize relevant insights and provide personalized summaries to help you decide whether it is right. This focus on synthesizing and adapting information will significantly improve user experiences, making it easier to digest large amounts of data in a meaningful way. These innovations will go beyond chatbots to incorporate more sophisticated AI applications across various industries.
AI regulation: Balancing innovation and oversight
AI regulation is a challenging and complex issue. While it's clear that AI is a powerful technology with significant consequences, regulating it effectively is complex.
Because of the complexities in regulation, I am not confident that new laws will drastically impact the pace of innovation or the market's direction. While AI regulation is an essential and challenging topic, it is unlikely to stop or dramatically alter how companies develop and implement AI soon.
The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/Andrey Suslov
Davor Bonaci, DataStax
Davor Bonaci is a proven software executive, joining DataStax following the merger with Kaskada, a venture-backed machine learning company, where he served as a co-founder and CEO. Previously, Davor served as the chair of the Apache Beam Project Management Committee and as an engineer at Google Cloud since its early days and the inception of Cloud Dataflow. Davor holds more than a half-dozen patents and earned his Master's degree from the University of Washington. Outside work, Davor is an avid sailor and can often be found racing on Lake Washington.